CENTER FOR DRUG EVALUATION ANDRESEARCH

APPLICATION NUMBER:

21-978

STATISTICAL REVIEWOO

US Department of Health and Human ServicesFood and Drug AdministrationCenter for Drug Evaluation and ResearchOffce of Translational Sciences

Offce of Biostatistics

STATISTICAL REVIEW AND EVALUATIONNEW DRUG ApPLICATION

NDA/Serial Number:

Drug Name:

Indication (s) :Applicant:

Dates:

Review Priority:

Biometrics Division:Statistics Reviewer:Concurring Reviewer:

Medical Division:Clinical Team:Project Manager:

Keywords:

CLINICAL STUDIES

21-978/N-000

Desonide Foam, 0.05%

Atopic dermatitis

Connetics Corporation

Submitted: November 28, 2005

PDUFA: September 21, 2006

Standard Review

Division of Biometrics III

Clara Kim, Ph.D.Mohamed Alosh, Ph.D.

Division of Dermatologic and Dental Products

Denise Cook, IVLD.jBindi Nikhar, M.D.

Melinda Harris- Bauerlien, NLS.

atopic dermatitis, desonide foam

Contents

1 EXECUTIVE SUMMARY

1. 1 Conclusions and Recommendations

1.2 Brief Overview of Clinical Studies

1.3 Statistical Issues and Findings

3

:3

3

:3

2 INTRODUCTION

2.1 Overview.

2.2 Data Sources

4Ll

G

3 STATISTICAL EVALUATION

3.1 Evaluation of Effcacy

3.1.1 Study Design .

3.1.2 Subject Disposition

3.1.3 Baseline and Demographic Da,ta .

3.1.4 Primary Effcacy Endpoints . . .3.1.4.1 ITT Analyses. . . . . .3.1.4.2 Sensitivity Analysis of the Primary Effcacy Endpoint.3.1.4.3 Per Protocol Analysis

3.1.5 Secondary Effcacy Endpoints

3.1.6 Effcacy Results over Time. .

3.1.7 Effcac)' Results by Center . .

3.2 Effcacy Results from Phase 2 Study (DES.C.202)

3.3 Evaluation of Safety ....

3.3.1 Adverse Events . . .3.3.2 Application Site Adverse Events.

5

G

5

8

9

10

10

11

12

13

15

Hi

17

18

18

18

4 FINDINGS IN SPECIAL/SUBGROUP POPULATIONS

4.1 Gender, Race, and Age. . .4.2 Other Special/Subgroup Populations

19

19

20

5 SUMMARY AND CONCLUSIONS

5.1 Statistical Issues and Collective Evidence

5.2 Conclusions and Recommendations

21

21

2:2

ApPENDIX 23

SIGNATURES/DISTRIBUTION LIST 24

NDA: 21-978 (Desonide Foam, 0.05%) ;3

1 EXECUTIVE SUMMARY

1.1 Conclusions and Recommendations

Desonide foam, 0.05% was statistically superior to its vehicle in a single study (DES.C.301) in

the treatment of atopic dermatitis. Effcacy was demonstrated after 4 weeks of treatment. AtWeek 4, 39% of Desonide subjects were successes versus 9% of vehicle subjects. The primary

endpoint and all secondary endpoints were statistically significant with p-values less than 0.0001

for each endpoint.

The adverse event rates were similar on Desonide and vehicle arms, where the vehicle arm's

rate was marginally higher than that of Desonide arm. The most common adverse event wasupper respiratory tract infection, and was reported by approximately 8% of the subjects. Thenext common adverse event was application site burning.

1.2 Brief Overview of Clinical Studies

The sponsor conducted one Phase 3, multi-center (17 sites in the United States) study (DES.C.301)

evaluating the safety and effcacy of Desonide Foam, 0.05% versus its vehicle in the treatmentof adolescent and pediatric subjects with mild to moderate atopic dermatitis. Subjects in thestudy were evaluated for up to 7 weeks, 4 weeks of treatment and 3 weeks of post-treatment

follow-up. A total of 581 subjects were randomized in a 2:1 ratio to receive Desonide Foam,

0.05% or vehicle. The primary endpoint was the proportion of subjects who had (i) Investiga-

tor's Static Global Assessment (ISGA) score of clear or almost clear (0 or 1) at Week 4 and aminimum improvement in the ISGA score of two grades from Baseline to Week 4; and (ii) score

of 0 or 1 for both erythema and induration/population at Week 4.The sponsor previously conducted a Phase 2 study (DES.C.202). A total of 106 subjects

were randomized in a 2:1 ratio to receive Desonide Foam 0.05% or vehicle. The primary andsecondary endpoints were identical to that of the Phase 3 study.

1.3 Statistical Issues and Findings

The sponsor has conducted a single study that demonstrated that Desonide foam, 0.05% is

superior to its vehicle in the treatment of atopic dermatitis, by a large margin. The studyresults were robust and consistent across investigational sites and subgroups. The sponsorconducted the study under the protocol that was agreed upon vvith the Agency in terms of

study design and primary endpoints. The Agency agreed at the Pre-IND /End of Phase 2 meeting

(3/30/2004) that one acceptable pathway would be to demonstrate effcacy in a single robust.highly statistically significant study. The primary endpoint was the proportion of success, where

success was defined based on the ISGA, erythema, and induration/papular.ion scores at \Veek

NDA: 21-978 (Desonide Foam, 0.05%) Ll

4. The difference in the success rates were strongly statistically significant (p-value~O.0001).

The protocol defined three secondary endpoints that were based on sum of symptoms score,

pruritus score, and ISGA score. Desonide was superior to the vehicle for all three secondaryendpoints. Subjects were followed for 3 weeks post-treatment. The Desonide arm success rate

decreased considerably after the treatment period ended at Week 4, whereas the vehicle success

rate continued to increase marginally post-treatment.

2 INTRODUCTION

2.1 Overview

Desonide is a synthetic corticosteroid. Desonide at a concentration of 0.05% is the active in-gredient of 10 FDA-approved drugs (Tridesilon, DesOwen, and several generics) for topicalapplication in three dosage forms: cream, ointment, and lotion. The sponsor obtained the right

of reference to NDAs 17-010 and 17-426, Tridesilon Cream and Tridesilon Ointment, currently

marketed by Clay Park Labs, Bronx, New York. The sponsor referenced the nonclinical andclinical safety data filed in NDAs 17-010 and 17-426 in support of Desonide Foam. In the current

application of Desonide Foam, 0.05% , the sponsor is seeking an indication of mild to moderateatopic dermatitis in patients from 3 months to 17 years of age, applied topically twice daily for

four weeks.

The protocol for the Phase 3 study (DES.C.301) was discussed in an End of Phase 2 meeting

held on .l1arch 30, 2004 and was evaluated as Special Protocol Assessment (SPA) in June 2004.

At the End of Phase 2 meèting, the sponsor was advised that three Phase 3 pathways could befollowed, which were

1. two independent, double-blind, vehicle controlled studies,

2. one 3-arm study demonstrating superiority of Desonicle foam to vehicle and non-inferiority

to an approved Desonide comparator,

3. one very persuasive, robust, double-blind, vehicle controlled study that is highly statisti-cally significant with no major flaws and consistent results across centers and subgroups.

The 'one very persuasive study' option was given primarily because Desonide Foam, 0.05% had

the same strength (0.05%), application route (topical) and frequency (twice daily) as previously

FDA-approved Desonide products. The sponsor submitted a single Phase 3 study as SPA, which

reflected the selection of the third option given above. Through the meeting and SPA review,

the sponsor and the Division carne to an agreement on endpoints and ~lost aspects of the study

design. The sponsor also conducted a Phase 2 study (DES.C.202) to get initial treatment effectestimates. Table i lists the clinical study programs. This review evaluates the Phase 3 effcacyand safety studies and briefly summarizes results from the Phase 2 study.

NDA: 21-978 (Des'cmide Foam, 0.05%) o

Ta.ble 1: Clinial St.udy Prograin

Number of 811 bjectsStudy Type

Dcsoniclc Vehicle 'Töta!

DES.C.202 Phase 2 72 ~i4 i (Ji

DES.C.:301 Phase ;3 :387 194 581

2.2 Data Sources

This reviewer evaluated the sponsor's clinical study reports and clinical summaries, as well as the

proposed labeling. This submission was submitted in CTD format and was entirely electronic.

The datasets used in this review are archived at

\ \ Cdsesub1 \n21978\N_000\2005-1 1-18\m5\53-clin-stud-rep \537-crf,.ipl\crt\datasets\301 and

\ \ Cdsesubl \n21978\N _000\2005- 11 -18\m5\ 53-clin-stud-rep \537-crf-ipl\ crt \ datasets \202.

3 STATISTICAL EVALUATION

3.1 Evaluation of Effcacy

3.1.1 Study Design

The sponsor conducted a Phase 3 study (DES.C.301) to evaluate the safety and effcacy of Des-

onide Foam, 0.05% in the treatment of mild to moderate atopic dermatitis. Protocol DES.C.301

was evaluated as a Special Protocol Assessment in June 2004. This study was designed as mul-

ticenter, double blind, vehicle controlled, and randomized. The treatment duration was 4 weeks

wit.h a 3 week follow-up period. The study was conducted in the United States between August.

2004 and June 2005. The study planned to enroll a total of 570 subjects from approximately 10

to 15 centers. The actual study enrollment was 581 subjects from 17 centers. The submission

ent.ry criteria included subjects from 3 months to 17 years of age, with mild to moderate atopic

dermatitis assessed by Investigat.or's Static Global Assessment (ISGA) (score 2 or 3), whosesum of score for erythema, induration/papulation, and oozing/crusting was greater or equal to

4, and involvement body surface area (%BSA) was greater or equal t.o 5%.

The enrolled subjects were randomly assigned in a 2:1 ratio to Desonide Foam, 0.05% and ve-

hicle groups. The randomization resulted in 387 and 194 subjects for Desonide Foam, 0.05% and

vehicle arms, respectively. The subject caregivers were instructed to apply the smallest amount

of medication t.o just cover all areas affected by atopic dermatitis and to apply not more thantwice daily for 4 weeks. Clinical evaluations were conducted at Baseline, Weeks 2, 4, and 7.

NDA: 21-978 (Desonide Foam, (L05%) 6

Week 4 was the primary time point for effcacy assessment.

For effcacy evaluation, the following endpoints were specified in the protocol.

. Primary: Proportion of subjects who have

ISGA score of clear or almost clear (0 or 1) at Week 4 and a minimum improvement

of 2 grades from Baseline to Week 4 (or end of treatment), and

a score of 0 or 1 for erythema at Week 4, and

a score of 0 or 1 for induration/papulation at Week 4.

. Secondary:

1'v1ean percent reduction in the sum of scores of erythema, induration/papulation,

lichenification, and scaling from Baseline to Week 4 (or end of treatment)

- The proportion of subjects who have a pruritus score of 0 at Week 4 (or end oftreatment)

The proportion of subjects who have an ISGA of 0 or 1 at Week 4 (or end of treatment)

and a minimum improvement in the ISGA score of 2 grades from baseline to Week 4

(or end of treatment).

The sponsor denoted the first secondary endpoint as the principal secondary endpoint and

added oozing/crusting to the sum of scores to be evaluated in the submission. This change waspartially in response to Agency comments and was done via a protocol amendment dated July

9, 2004, which was before the first subject was enrolled (8/31/04). The scoring systems usedto assess effcacy in the primary endpoint (ISGA, erythema, and induration/papulation scores)were based on a 5-point scale, defined as the following.

ISGA:

0 Clear:

1 Almost Clear:

2 Mild:

3 Moderate:

4 Severe:

l\Iinor residual discoloration; no erythema or indura-tion / papulation, no oozing/crusting

Trace faint pick erythenia, with almost no induration/papulation,and no oozing/crusting

Faint pinl\: erythema" with mild induration/papulation and no ooz-

ing/ crusting

Pink-red erythema with moderate induration/papulation and some

oozing/ crusting

Deep or bright red erythema with severe induration/papulationwi th o07:ing/ crusting

NDA: 21-mS (Desonide Foam, O.05(Yr)) 7

Erythema:

o Absent:

1 Minimal:

2 Mild:

3 Moderate:

4 Severe:

No erythema present (may be minor discoloration)

Faint pink, barely apparent

Pink-red, noticeable

Pink-red, easily noticeable

Deep or bright red, may feel warm to the touch

Ind uration /Papulation:

o Absent:

1 Minimal:

2 Mild:

3 Moderate:

4 Severe:

No evidence of elevation

Barely perceptible elevation

Perceptible but not extensive elevation

Marked and somewhat extensive elevation

Marked and extensive elevation.

The protocol and submission defined the intent-to-treat (ITT) population as all subjects who

were randomized and received the study drug. Subjects were excluded from the per-protocol

(PP) population if they missed more than a total of 10 applications at any time or six consecutiveapplications of the study medication, or did not have effcacy evaluations at Baseline and 'Week

4 visits, or used prohibited medications at any time during the treatment period. The ITT and

PP populations were analyzed for effcacy, where ITT analysis was the primary.

The analysis methods proposed in the protocol and submission were the following.

. The success rate was analyzed with the Cochran-Mantel-Haenszel (CMH) test stratified by

investigational center. The center by treatment interaction was tested using the Breslow-Day test at a significance level of O. i.

. The principal secondary endpoint, the mean percent reduction in the sum of scores analysis

was based on an ANOYA model with tenus for treatment, center, and treatment by site

interaction.

. The other secondary endpoints (proportion of subjects with pruritus score of 0; and pro-

portions of subjects with ISGA score of 0 or 1 with a minimiun improvement of 2 grades)

were analyzed using CMH test, stratified by center.

. Last-Observation-Carry-Forward (LOCF) was used to impute missing values in the eff-

cacy endpoint. In addition to the LOCF itpproach, a sensitivity analysis based on multi-ple imputations using sequential generalized logistic models. Separate generalized logisticmodels were specified for each visit that had missing data. Logistic: models for the de-

pendent variable with the least missing data was f;:òtiniated Cirst ,lond proceeded to the

dependent variable with the most missing data.

NDA: 21-978 (Desonide Foam, 0.05%) S.,'-

. The target number of subjects per site was at least 30 subjects (20 in the Desonide Foam

group and 10 in the vehicle foam group). Sites with smaller enrollment were combined

based on geographical location and climate similarities. According to the protocol, com-bination of the investigative sites was completed prior to unblinding the data.

Study sites were pooled into 11 investigational groups, which were used in the effcacy

analyses. Table 2 presents the pooled sites and the number of subjects in each site before

pooling.

Table:2: Enrollment by pooled sites

Number of Number ofPooled Site Original Site

Siihj(~ct;.Pooled Site Original Site

Subject;.

1. 5

2"'J

:\ 4

.1 (;

5 7

G 8, j IJ

:32, 12

n 5024, 19, 13v,

7 9

8 lJ

0 12, 1:\. 16

10 1.1

11 2, 15. 17

;) I

.j,!

GO

2:3,28

5::)

21,11. 8

Source: Reviewcr analysis

3.1.2 Subject Disposition

The study enrolled a total of 518 subjects from 17 study sites and randomized them in a 2:1ratio to treatment and vehicle arms. Thus, 387 subjects were randomized to Desonide arm and

194 to the vehicle arm. Table 3 presents the reasons for study discontinuation.

Table 3: Beason for Study Discontinuation

J)csoiiidc

N,o':38í

Vehiclc

N=Hj.j

Subjects who discontinued!lca.soil

Adyersc Event 2 (51.7c)Noii-Compliance (j (2%)

33 (9%) 54 (28%)

Discnsc ProgresC'ioli 3 í 1 e¡;,)Subject. R(;quest to \Vithdr,m- :3 (lX;)D('Htll I (-(:1 if.)Other 18 (5%)

17 (9%)

:3(2':';,)

IS (!Y7¡,)

12 (6%)

o (OS;)

.1 (2%)

.source: Sr.udy rcport ":)CJ! 2-IJe.s-c-:\1J1.¡idf. pg. 41

NDA: 21-978 (Desonide Foam, 0.05%) 9

The number of subjects who discontinued the study was higher in the vehicle arm than

the Desonide arm, at 33 (9%) and 54 (28%) for Desonide and vehicle arms, respectively. The

most common reason for study discontinuation in the Desonide arm was 'other'. Out of the 18

patients whose reason for discontinuation wa.'ì 'other', 1 subject misunderstood the visit time

lines, 1 subject was 4 weeks late for Visit 3, and the remaining 16 were lost to follow-up. For the

vehicle arm, disease progression (18 subjects) and adverse events (17 subjects) were the most

common reasons for discontinuation. The reasons for the 3 subjects request for discontinuation

in the Desonide arm were worsening of condition, inconvenience, and subject not being able to

make visits. In the vehicle arm, 3 subjects requested to withdraw due to worsening of condition,

7 subjects due to lack of effcacy, and 1 subject due to lack of time.

3.1.3 Baseline and Demographic Data

Table 4 presents the Baseline demographic data. The Baseline demographic variables were

generally balanced across treatment arms. The average age of the subjects was approximately

6.9 years and the age range was from 3.6 months to 18 years. Desonide arm had slightly moremale subjects (51 %) than females, whereas the vehicle arm has more females subjects (54%) than

males. Approximately half of the subjects were Caucasian on both arms. Race was relativelybalanced across treatment arms.

Table 4: Baseline Demographic Data

N=:387

Desoiiide Foam Vehicle

N=194

Age(iii years)

llj('¡l! (std)

iicdianlIiili,TlIIX

7D (4.8)

C.2

(0.:3, S.O)

6.8 (UJ)

5.1

(D. 1.IS.O)

Genderl\fakFemale

198 (51%)

LXr¡ ('19%)

90 ('W'!)

10,1 (Cd ';¡ )

RaceC~a\lcasinn L rn ('lD';i;)

9J (2-1'/;)

66 (J 7~!r;)

17 (l?,O

HI (S%)

A hicaIl-A lIH'ri Cil I

llispanic

Asian

Other

ion (S2";¿,)

Hl (25';,\',)

32 (WI!\)

G (:3%)

7 (,I;;i)

Source: Stiii).- I"'JH1rt. C,:3512.c!cs-c.301.pclL )))2.,11

Table 5 presents the Baseline ISGA scores and the extent of atopic dermatitis (%BSA). The

NDA: 21-978 (Desonide Foam, 0.cí5%) 10

Baseline severity scores, the ISCA score and Body surface area (%BSA), were fairly balanced

between the Desonide foam and vehicle arms. Desonide arm had a slightly higher proportionof subjects with moderate severity and a marginally higher mean %BSA score than the vehicle

arm. Although the study population was mild to moderate atopic dermatitis, one subject withISCA score of 4 was enrolled in the vehicle arm. All subjects had atopic dermatitis on at least

5% of their body surface. The median%BSA was 15% and 13%, and the maximum surface areawas 97% and 90% for Desonide and vehicle arms, respectively.

Table 5: Baseline Severity

Desoiiiclc Fo;ini

N=:387

Vehicle

N=l~H

Iuvcfòtigator's Static Global Assesfòuicnt Score2

~)

145 (:¡7'/n)

242 (ti:l/,;)

o (OYi,)

71 t:lSfV¡)

119 (61%)

1 (1 'rr,)I

Extent of Atopic Dermatitis (%BSA)mCill (std)

iiicdinii

niiri,nu\x

21 (187)Ui

(5,97)

19S (175)

1:3

(5,90)

Source: Study rcport 53512-clcs-c-30!. pelf. pg.4:5

3.1.4 Primary Effcacy Endpoints

3.1.4.1 ITT Analyses

The protocol defined success as subjects who had

. ISCA score of 0 or i at 'Week 4, with a iniiiimum improvement of 2 grades from Baseline

to Week.1; and

. Erythema score of () or 1 at \Veek 4; and

. Induration/papulation score of 0 or 1 at vVeek 4.

Table 6 presents the primary effcacy results in the intent-to-treat (ITT) population. AtV'¡eek 4, the primary time point, 39% of the Desonide foam subjects reached success status,

,vhile 9% of the vehicle arm were successes. The difference of the success rates in the twoarms was highly statistically significant with a ¡i-valuc of ~ 0.0001, establishing the effcacy of

Desonide foam at Week 4

Since the primary endpoint is clefiiied as success only when several conditions are met, this

reviewer compared the success rate of each component of the primary endpoint. Table 7 presents

NDA: 21-978 (De::onide Foam, 0.05%) 11

the success rates of each component of the primary endpoint. Differences between the Desonide

and treatment arms in all three components of the primary endpoint were highly statistically

significant with p-values less than 0.0001.

Table 6: Pliniary EfIcacy Endpoint Results (ITT)

Dewnidc Foam Vehicle

N=:387 N=HJ4 P-\-'¡Illt

Success 152 (39(X) 18 ( 9(/';) .(:0.0001

p-vciliies arc calculated usirig Cl\HI sl.,üistic stnil.ificd by pool cd sites

Soiirc(~: Stiidy report ,)3)12-des-c-:30 1.pdL pg. '",1

-fable 7: Compoiicriis of !'rinlèirv Endpoint R.esult.s

Dcsoi 1 i, i c Foam Vcliicle

N=:187 N'=l()4p-niliw *

ISGA t 1:..)7 (11

2G2 (GS';;,)

2G8 (G9C;;)

J8 ((i';.n

G9 (:3()':,O

0(. O.OOOJ

.( 0.00010(. G.OGOl

Er.vtlieiia t

Induration/PapulatiOlI ¡ 7:3 (38~;~)

. p-valui's arc calculMcd using CivJH st,itist.ic "trahfied by pooled sitesL Success is dcfîiieci as score of 0 or 1 I.i\: \Vcek ,.1 with a iiinimmii

iinprovciilCrit of 2 gradcs from J3ascliiie to \Vce). 4

:t Success is ddiiJ(d ,is SC01"e of () PI i i' \: \V,'(~k ,i

Source: n.cvicwcr iJi leJ Ivsis

3.1.4.2 Sensiti"it.y A.nalysis of the Primaq Effcacy Ene/point

Per protocol, last observation carried forward (LOCI") was used to impute missing data in

the analyses of the previous section. This reviewer conducted two sensitivity analyses to ensure

that the effcacy results were not driven by the imputation method. The first analysis imputedall missing observations as successes for both arms and the second analysis imputed missingdata as failures. Table 8 presents primary effcacy results using these imputation methods.

The number of missing subjects at Week 4 wac; a total of 73 observations (12.6%), 28 (7%) and

45 (23%) in Desonide and vehicle arms.. rec;peclivdv. Thus, the vehicle ann had a larger number

of drop-outs than the Desonide arm. Imputing the mi::sing data as successes is a conservativeapproach, since the vehicle arm has more niissing data and the most common reason for drop-

out was due to lack of effcacy, while that of the Dcsonide arm was lost to follmv-up. However,

NDA: 21-978 (Desonide Foam, 0.05%) 12

Table 8: Sensitivity Analyses on Primary Endpoint

DeRonide Foam Vehiclc

N=387 N==191p-yahie'

Imputed subjecht

Sllccess t

28 ( 7'lt)

179 (-W%)

151 (:39e¡;,)

-1:5 (23':;¡,)

62 (;2Sn

17 ( 9%)

f). DUO,:

'C: 0.0001SU(:ccss 1"t

p-valueR arc calculated using CIVrH statistic stnnined by pooled sites

.,. Nllinber of missing suhjects at 'Neck .J

T l\E%iiig data imputed as succcsses

iT Missing dat" iinputed as failiircs

Source: neviewcr iJmilysis

even using this approach, the difference in proport.ion of success in the two arms was stat.istically

significant with a p-value of 0.0009, in favor of the treatment arm. The second analysis (imputing

missing as failure) results are similar that using LOCF. The sensitivity analyses ensures thatthe statistically significant results was not driven by using LOCF as the imputation method.

3.1.4.3 Per Protocol Analysis

The per protocol (PP) population excluded subjects who had missed more than a total of 10applications at any time or 6 consecutive applications of study medication, subjects who did

not have effcacy evaluations at the Baseline and Week 4 visits, and subjects who have usedprohibited medications at any time during the treatment period. A total of 100 subjects (17%)

were excluded from the PP population, 41 subjects (11%) and 59 subjects (30%) in the Desonide

and vehicle arms, respectively. The most common reason for exclusion in the vehicle arm was

due to the subject missing more than a total of 10 applications (51 subject.s) followed by no

effcacy assessments at Baseline or vVeek 4 (6 subjects). For the Desonide arm, the subjects

were excluded due to the two reasons above with equal frequency (18 subjects) Not.e t.hat some

subjects were excluded for multiple reasons. Table 9 presents the results of the primary endpoint

analysis at vVeek 4 based on the per protocol population.

The proportion of successes in the per protocol population was higher for bot.h arms thanthe ITT population at 42% and 13%, Desonide and vehicle arms, respectively. The difference in

the proportion of success of the t.vo arms was strongly st.atistically significant with a p-value less

than 0.0001. The ITT and PP population primary endpoint analyses results were similar, which

further supports the superiority of Desonide foam over vehicle. The analyses results of eachcomponent of the primary endpoint based on the PP population can be found in the Appendix

(Table 17).

NDA: 21-978 (Desonide Foam, 0.05%) 13

Table 9: Primary EfIcacy Encli:ioint Results (PP)

Dcsoiiide Vehicle

N=:)46 N=I:35p-value

Siiccc~~ 147 (,l2'XiJ 17 ( B%) -cO.()OOI

P-VH luc~ arc calculated u~ing C\'IH ~tatistíc ~tratificd by pooled sites

Source: Study report 5:3512-des-c-301.pclf, pg. 6:3

3.1.5 Secondary Effcacy Endpoints

The secondary endpoints defined in the protocol were

. Mean percent reduction in the sum of scores of erythema, induration/papulation, licheni-

fication, and scaling from Baseline to Week 4 (or end of treatment)

. The proportion of subjects who have a pruritus score of 0 at Week 4 (or end of treatment)

. The proportion of subjects who have an ISGA of 0 or 1 at Week 4 (or end of treatment)and a minimum improvement in the ISGA score ot 2 grades from baseline to Week 4 (orend of treatment).

The sponsor's estimation of the first secondary endpoint (Source: 53512-des-c-301.pdf, pg.

154) was based on a population that excluded 24 subjects from the ITT population, 18 and 6subjects from Desonide and vehicle arms, respectiyely. The rational of excluding these subjects

were not stated in the submission. The sponsor"; analysis resulted in mean percent reductions

in the sum of scores for clinical signs of 60.0% 'md 20.9% for Desonide foam and vehicle arms,

respectively. In this review, the first secondary (principal secondary) endpoint was reanalyzed

using the same population that was used lor the primary endpoint analysis, thus the ITTpopulation using LOCF to impute missing observations at \Neek 4. Table 10 presents theresults of the principal secondary endpoim using the ITT population.

This reviewer's analysis using the ITT population resulted in mean percent. reductions of

57.2% and 20.2% for Desonide and vr"hicle anus, respectively. These result.s are marginally

smaller than those of the sponsor for)oth arms. The difference in the mean percent reduction

from Baseline to \Neek 4 of the (\\"0 l.rms are st.atistically significant with a p-value of less than

0.0001.

The sponsor analyzed t.he additional secondary endpoints (the proportion of subjects wit.h

a pruritus score of 0 at V/eek 4: and the proportion of subjects with modifìed success at Week

4) based on the ITT popiihtiun which is present.ed in Table 11. Modified success was defìned

as subjects who had an ISG'i score of 0 or i at \Neek 4 (or end of t.reatment) and a minimum

improvement in t.he ISG:-. :ocores of 2 grades from Baseline to Week 4.

NDA: 21-978 (Desonide Foam, 0.05%) 14

Table 10: Mean Percent Reduction in the Sum of Scores for Clinical

Signs fron1 Baseline to Week -i (ITT)

Dc~onidcF'oarn Vehiclc

N=387 N=194¡¡-valiwf

Baselinc

ìVeek -¡¡End of Treatment

Percent Tlccluction from Baseline

9.G (2.8)"

4.1 (:3.8)

57.2 (37.5)

9.1 (2.7)

1.8 (4.6)

20.2 (/10.5) ,; 0.0001

;f'Numbers in parentheses represent the stanclDrd deviation

t p-valnc is derived from a paramctric ANOVA model with tcrms for

trcatmeiit and pooled sites

80u1"(e: Reviewer aTliJlysis

TnbJe J 1: Additionnl SecondnTY Endpoints (ITT)

Desoiiide FOiln Vehicle

N::=:387 N=l9-1p-vcllue '

Pruritns Score of n 13:3 (;4%)

157 (11 'Y¡,)

Hi (10%) -.o.ooni

:\lodif1cd Success f is (9'7c,) -.O.DOOl

"" p-value is derived freiin Coehrèln-lvlalltcl-Haeiiszel test stratified h),- pooled sites.

T l\Jodified Siicn~s:, is (lcfined H:' the proportion of suhjects ~wlio have aii ISGA

score of II or 1 at ìVeek ~i and a minimum improvemellt in the 18GA :,core of :2

grades from Baseline rei ìNeck 4.

Source: nevie\yer inialysis and Stiidv rqiDrt 5:3"Jl2-des-c-30 l.pdf. ¡¡g. 72 mid pg.

7i

Thirty four percent of the Desonide arm and 10% of the vehicle arm had a pruritus score of 0at ,Veek 4, using LOCF to impute the missing observations per protocol. One subject on thevehicle arm had a pruritus score of 0 at Baselíne, but did not have an assessment at Week 4.

The sponsor's analysis imputed the missingness as failure (score greater than 0). Thus, Table 11

shows one more score 0 in the vehicle arm than the sponsor's analysis. However, the discrepancy

does not. alter the conclusion that the difference between the two arms' proportion of subjects

who had a score of 0 in pruritus score is statistically significant, with a p-value less than O.OOOL

The difference of t.he proportion of subjects with modified success is also highly significant with

a p-\'ahie less t.han 0.0001. which was discussed as a component. ofthe primary endpoint in Table-I.

NDA: 21-978 (De::onide Foam, 0.05%) 15

3.1.6 Effcacy Results over Time

Subjects were followed for a total of 7 weeks, including a 3 week post-treatment follow-up period.

Subjects were evaluated at Baseline, Week 2, Week 4 (End of Treatment), and Week 7 (3 Week

post-treatment). This reviewer presents the primary endpoint. success rat.e over time in Figure

1. Along wit.h the success rate, unadjusted 95% confidence intervals are presented to show thevariability of the success rates at each visit.

Figure 1: Success over Lime

0.5

0.'I..~,.-.--.'--()esonideL.'-~,.:,;_-,'___~b.!e

0.3

0.:'

0.1 -

Visit (weeks)

The success rate of Desonide foam continued to steeply increase throughout. the treatmentperiod, but decreased considerably after the treatment period ended at Week 4. The success rate

of Desonide foam was 21.7% and 20.7% at Week 2 and Week 7, respectively. (Week 4 success

rate was 39.3%) indicating a lower response rate at 3 weeks po:-t-trcatment than 2 weeks into

the treatment (see Appendix on page 24). The vehicle ann success rate increa,'3ed throughout

the treatment period, however at a much smaller degree than the treatment arm, and continuedto increase marginally during the follow-up period. Although the Desonide arm success ratedecreased post-treatment, the success rate in the Desonide foam ann remained higher than that

of the vehicle arm at Week 7 at. 20.7% and 12.4%, Desonide and vehicle arms, respectively.The 95% confidence intervals of the success rates do not overlap during the treat.ment period,

Desonide group being superior to the vehicle. HovveveL the confidence intervals overlap slightly

at the post-treatment evaluat.ion.

Figure 2 presents the scores of the three eleiicrits that furm the succc:-s criteria (ISGA,

erythema, and induration/papulation) over time. Note that higher scores indicate more severe

NDA: 21-978 (Desoriïde Foam, 0.05%) 16

symptoms. The scores of the Desonide foam in all three elements continued to decrease until

the end of treatment and increased after that, which is consistent with the overall success rate.

The scores of the vehicle arm are also consistent to the overall success rate, which decreased at

a more marginal degree than the treatment arm but continued to decrease post-treatment.

Figure 2: lSGA, Erythenia and IndurationíPapulation scores over time

_______~.._LISGA

- 2).,

DesonideVehicle

~:.O

l.~;

'"8U)

1.11

E hem"s Induration/Pa ufation

2.!'

1.:';

2 n

"-., .1.0

VIsit (Weeks)

3.1.7 Effcacy Results by Center

This study involved 17 investigators, all from the United States. Each investigator enrolled

betvveen 8 to 60 subjects. Sites were pooled based on geographical and climate similarities if

enrollment was less than 30 subjects per protocol (20 in Desonide foam group and 10 in vehicle

arm). According to the sponsor 1 pooling was done before unblinding of the data. There were11 pooled irivestigative sites.

This reviewer presented the success rate and number of subjects enrolled in each pooled site

by treatment in Figure 3. The success rates of both arms appeared to be relatively consistentacross pooled sites, and therefore the results do not seem to be driven by extreme sites. TheBreslow-Day test results also supported this conclusion with a p-value of 0.6388.

NDA: 21-m8 (Desonide Foam, 0.05%) 17

Figure :3: Effcacy by pooled sites

~I. Desn",,;"

= IcO

. 33

. 29 . 47

. 29

W -:i-~ 0~w~ü5

. 29 9,40

. 3"/

iJ .38

.30 . 38NcO

',~ ( ¡" 217

¡ ':.~

=cO

r '" "" - -- ~----~~._-- TO' .'........ .'-."'T -- '. ".'.'..''''-r ... ----T.---..-..--.----.-r-~--..--.. '-T -'---.~.

Pouled site

3.2 Effcacy Results from Phase 2 Study (DES.C.202)

The sponsor conducted a small Phase 2 study that randomized 106 subjects to Desonide Foam,

0.05% or vehicle group in a 2: 1 ratio, 72 and 34 subjects for Desonide and vehicle arms, respec-

tively. The primary and secondary endpoints were identical to the Phase 3 study. The subjectinclusion criteria, (ISGA score of 2 or 3; a sum of scores for erythema, induration/papulation,and oozing/crusting of 2' 4; and an involvement of 2' 5% total body surface area) were alsoidentical to the Phase 3 study. Table 12 presents the point estimates of the primary endpointanalysis from the Phase 2 triaL.

'Libk 12: l'rinJaryEndpoinl Analysis (Phase 2)

Dc~onidc h'dlll

N72Vcliiclc

N=3.j

SUCC:C,:,:i 3D ( J (1:2%)

, Success was ddìricd idciiticall.\' 1;0 the primary

Of.tlj(I' Phase' ;1~1!1(1\'.

Sul)nc: Stwly l'~p"rt !/)5 J i -d"':-"-2fl.pdL pg, ')2

The success rate in the Desonicie ann (54%) was much greater than that of the vehicle arm

(12%) and also larger than the Desonide arm success rate (39%) of the Phase :1 study. Although

NDA: 21-978 (Desonide Foam, 0.0.5%) 19

Ta.ble 14: Application Site Adverse Events

Application site AEsConditions

Atrophy

Buniing

DcnnatitisDcsqllmriatioii

Ervtlicina

Pigllltiitation cliaiiges

Hmctioii

Urticaria

DCSOTlic1c Vehicle Total

N=~\S7 n=1£)/1 N=581

22 (ß%) 27 (14%) 49 (8%)

;OJ (1 )¡,) 0 (0%) 5 (J%)J 1 (3'7.) IS (8%) 26 (1%)2 (1%1 1 11%) :3 (1 ')~,)ii íO'ì,,) 2 (1%) 2 (o~nI (0';7() :3 (2'ì(j) 1 (i~n1 (0%) 2 0%) .) (1'7,))oJJ (1%) G (:3S;;, ) D (27,,)

0 (0';1)) 2 ll;U 2 (0%)

Source: Stucly report 5:3:J12-,!cs-c-:H)J. pg. 101

The most common adverse event concerning application site was application site burning. Alarger proportion of subjects on the vehicle arm (8%) experienced this event than the Desonide

arm (3%). With the exception of atrophy, which .5 subjects on the Desonide arm and noneon the vehicle arm experienced, more vehicle subjects experienced local adverse events thanDesonide subjects.

4 FINDINGS IN SPECIAL/SUBGROUP POPULATIONS

4. i Gender, Race, and AgeTable 15 present.s t.he success rates by gender, race and a.ge groups based on t.he ITT populat.ion.

Effcacy does not appear to be affect.ed by gender, race, and age. Male and female subjects in the

t.reatment arm had similar success rat.es, which was also consist.ent across age cat.egories. The

success rate in the Hispanic group was slight.ly higher t.han that. of other groups. Nevert.heless,

t.he subgroup analysis supports the claim t.hat the Desonide arm success rat.e is superior to that

of the vehicle arm across subgroups.

NDA: 21-978 (Desonidf' Foam, ().05%) 20

Table 15: Subgroup Analysis (ITT)

l)esoiiide FÜr.l.lll \'ehiclt'

N:-=::iB7 N=194

Total 19B uo"laIc

Succc:;:: (';;~ì 7b r :19~'~,) 12 (1:3(;;(',)_.._-----GenderlDtal 11'9

lOO

FemaleSucccs~ (C:'r,) 7J (:~!)~+)

Tot-a.! 1:J1Ciiiica..,:in,n

Suee"" (%)

TotalfY~) (:J.J'I(:l

!ll;\ j'ricai J.- Arnerica,n

Sl1C(,(,:S~ :?7 (~!YX)

104

6 ( 6';;)

.S (:)(ì~,)

4!1

(j

Race'I"tal (-() T2

IIi:pal1Î('SUC(,t)S~ (':v~) I!) Itn ('~i

Total :'H.iOt.lier

Succ(-~s:- ((7i~) 211 \;,fi'::';.:)

T'ohil 7t;

il2 Yl'~. ¡,~ yr;:qSUCCC!:;- (%:)

T"Ui.:2~) (:3t)1,')

in~6 .\T:". 1'2 .yr~)

Success (%)

4 (1:3%)

i.'.)

:1 ¡2:YX)

:31

1 (12(';,;co',

'rotaJ ,'iG

,Ix (:1~Jl!'_,) 7 (1:1(~~(ì----~------"-------_.~17

Age

(:Jyr", (j yrs)Succe:-s ('~j)

'¡"tal:3;; 1."11%)

102

i:3 ritJl:-:L :3 yr,s)Success (i¡~,.ì lO (:NYi.) i ( 7C~)

Soiicc: Stiid~' report 5:3512-des-c-:301. pdf, pg.J9-G i

4.2 Other Special/Subgroup Populations

:3 ( 6'Yr,)

GO

The proportion of success rate was explored by Baseline disease severity. The sponsor analyzed

success rate at vVeek 4 by Baseline ISGA score. This reviewer included an additional analysisof success rate by the extent of atopic dermatitis (%BSA) at Baseline. The subjects werecategorized into two groups, %BSAS; 14 or %BSA:: 14, where 14 was the median %BSA of all

subjects. Table 16 presents the success rate by baseline severity.

Success rates were slightly higher for subjects with moderate atopic dermatitis (ISGA 3) atBaseline than mild subjects (ISGA 2). This may be due to the fact that subjects vvith mild

atopic dermatitis had to reach clear, while moderate subjects could reach clear or almost clear to

achieve success status. The success rate in subjects with %138A below the median was slightly

higher than that of subjects with %BSA above the mediai!.

NDA: 21-978 (Desonide Foam, 0.05%) 21

Table 16: Subgroup Analysis: Baseline Disease Severity

Ba~dii¡(' rSGA

De~onide Föani Vehicle

N==~\87 N=1~H

Total IDS 902

Success ('K) H (:\0%) :3 ( l'7c)

Total 2-12 120t:3

Success lU8 (J;)%) 15 ( i:3';¡r,)

Tùtal 190 102.::: 11

Success (%) 87 (LGï\; 10 (10%)

Total 197 92~. 14

D'!(i)Success (1/,,) 65 (:3:3%) R (c.

Ba~dilJe '/iBSA

T '11)(' :òubjcct with ha,s()iiuc rSGA SCf)J"' of -1 wn~ included in ISGA ::\

Source: Study report:J:1512-de:ò-c-:301. pel t, pg.G2 aiid reviewer iJiiaiysb

5 SUMMARY AND CONCLUSIONS

5.1 Statistical Issues and Collective Evidence

The sponsor submitted the results of one pivotal Phase 3 trial and one Phase 2 trial to supportthe effcacy and safety claim for Desonide foam, 0.05% in the treatment of atopic dermatitis.The sponsor has conducted a single Phase 3 study that demonstrated that Desonide foam,

0.05% is superior to its vehicle in the treatment of atopic dermatitis, by a large margin. The

study results were robust and consistent across investigational sites and subgroups. The sponsor

conducted the study under the protocol that was agreed upon \'iith the Agency in terms of study

design and primary endpoints. The Agency agreed at the Pre- IND /End of Phase 2 meeting

(3/30/2004) that one acceptable pathway would be to demonstrate effcacy in a single robust,highly statistically significant study. The primary endpoint was the proportion of success, where

success \-vas defined based on the ISGA, erythema, and induration/papulation scores at \Veek

4. The difference in the success rates were strongly statistically significant (p-value-cO.OOOl)

The protocol defined three secondary endpoints that were based on sum of symptoms score,pruritus score, and ISGA score. Desonide was superior to the vehicle for all three secondaryendpoints. Subjects were followed for 3 weeks post-treatment. The Desonide arm success ratedecreased considerably after the treatment period ended at \Veek 4, whereas the vehicle success

rate continued to increase marginally post-treatment.

Although the Phase 2 trial was not powered for statistical inference, the results were consis-

tent to the findings from the Phase 3 study. The success rate in the Desonide arm was much

NDA: 21-978 (Desonide Foam.. 0.05%) 22

greater than that of the vehicle arm.

5.2 Conclusions and Recommendations

Desonide foam, 0.05% was statistically superior to its vehicle in a single study (DES.C.301)

in the treatment of atopic dermatitis. The results from this Phase 3 study were consistent to

those of the previously conducted Phase 2 triaL. Effcacy was demonstrated after 4 weeks oftreatment. At .Week 4, .39% of Desonide subjects were successes versus 9% of vehicle subjects.

The primary endpoint and all secondary endpoints were statistically significant with p-valuesless than 0.0001 for each endpoint.

The adverse event rates were sirnilar on Desonide and vehicle arms, where the vehicle arm's

rate was marginally higher than that of Desonide arm. The most common adverse event wasupper respiratory tract infection, and was reported by approximately 8% of the subjects. The

next common adverse event was application site burning.

A~tAIlSl:

W Olllf\ -"1$ llA""J/nlAl '

NDA: 21-978 (Desonide Foam, 0.05%) 2:

ApPENDIX

Per Protocol Primary Endpoint Components

Table 1ï: Components of Priniar.y Endpoints (PP)

Desonide Foain Vehicle

N=346 N=J:5p-vahie'

ISGAt

Erythema t

Ind uration/ Papulation:'

152 (MX)

251 (7:l/c)

2ö9 (7ö'X)

17 (1:3Sn

GO (41'7,,)

G6 (4D'!1)

.( G.OGOl

.: 00001

. p-values are calculated using C~J\UJ statistic: stratified by poolcd sites

t Success is defincrl '-:' scorc of U or 1 at \Vcck .1 with n iniriin.inrii

improvement of 2 grades from Baseliiie tu \Vcck

t Success is defined ,is score of G or 1 ,it \Veek .j

Source: Reviewer ¡Dwtvsis

APPEARS TH

ON ORIG/~~LWAY

NDA: 21-978 (Desonide Foam, 0.05%) 24

Table of Effcacy Over Time

Table is: TiJble of Efficacy Ovpr Time

DesolJidc Foain Vdüdc\Vcck

N=::l87 N=I();I

o

2

o ( O'x,)

8l (21.%)

152 (:m:)';o

80 (20.7%)

() L 0%)7 ( :Uj'ï.)

18 (!L!%)24 \l2A%1

1

ï

Sourcc: Reviewer analysis

SIGNATURES/DISTRIBUTION LIST

Primary Statistical Reviewer: Clara Kim, Ph.D.

Date: June 19, 2006

Statistical Team Leader: Mohamed Alosh, Ph.D.

cc:

Archival NDA

DDDP /Walker

DDDP /Kukich

DDDP/Luke

DDDP /Lindstrom

DDDP /Nikhar

DDDP/Cook

DDDP /Harris-Bauerlien

OBIO/O'Neill

OBIO /Patrician

DBIII/WilsOii

DBIII/Alosli

DBIII/Kim

June 19. 2006

----------------------------------------------.-------...._._-------_._----_._---------........_..._-_.-._-----------This is a representation of an eleçtronic record that was signed electronically andthis page is the manifestation of the electronic signature.-------------------------------_._---------------------------------------------------------~-------------------------/s/

Clara Kim6 /2 1/2 0 0 6 02: 04 : 44 PMBIOMETRICS

Mohamed Alosh6 /2 1/2 0 0 6 04: 21 : 55 PMBIOMETRICSConcur with review